[c-nsp] Cisco BFD and NSF

[Charles Spurgeon]

I didn't have a TAC case open and therefore didn't nail down a formal
bugID. That's my fault, I suppose, for working with someone at Cisco
who I've worked with for years and using them as a "backchannel" to
communicate the issues I was seeing with BFD to Cisco DEs.

Instead, I got a report that the developers would revisit the
pre-emption problem and also work on providing an alert/log in cases
where they detected that BFD packets were being lost to
pre-emption. Longer term I was told that they were looking at giving
BFD better priority in IOS. This activity was as of late April, '06.

-Charles

On Mon, Jul 31, 2006 at 09:17:03AM +0300, Saku Ytti wrote:
> On (2006-07-30 11:37 -0500), Charles Spurgeon wrote:
>  
> > Apparently they are having problems with BFD access to the CPU during
> > various file writing events. Cisco is aware of the issue and I was
> > told that there were bugfixes in the pipeline. IIRC, the fixes were
> > committed to the 7600/12.2SR branch. 
> 
> What bug ID's? To my case (602003831) TAC insisted (and I insisted
> opposite) that this is not a bug, and I just have to configure less
> aggressive timers. However this was almost a year ago in 12.2(18)SXE3
> perhaps amount of complains have made them change their mind.
>  My timers were 50ms by 5 packets, which hardly can be considered
> aggressive in BFD world. 
>  Initially TAC did suggest us to tune 'process-max-time', so that
> bg. BGP would be suspended early, we did this, but even after
> minimum 'process-max-time' we could observe BFD flap.
>  One thing I did notice from 'show process cpu' was that BFD
> process isn't ran as high priority as it could be, but this probably
> would be wrong place to fix the issue.
> 
> Thanks,
> -- 
>   ++ytti
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/